Jump to content
HWBOT Community Forums

Xtreme Addict

Members
  • Posts

    2403
  • Joined

  • Last visited

  • Days Won

    2

Everything posted by Xtreme Addict

  1. Roman's is a good one. I think I would just replace benchmarks with games fps, it will bring more fresh blood to HWBOT and we XOC guys are already extinct species :banana:
  2. Our fight with making Titan X Pascal @ Epower was amazing and one of the most epic challenges ever.
  3. You had better efficiency @ clocks, I don't know how you managed to get that good run! but my CPU did some crazy clocks with pretty okay efficiency, good enough
  4. Normally I don't comment other's threads sales. 1st thing: you say 6520 MHz @ 1.92v, you show 6500 @ 1.94v. Other cpus don't have screenshots, not even talking about HWBOT Prime. From the past I remember you were selling many cpus with same screenshot. 2nd thing: you say BOX, but sold as OEM, decide, but from what I understand this is tray CPU (OEM) with same batch as BOX. We binned with Sam our asses off during summer break with Vietnam batches, and we found many good cpus, but couldn't find @ 6600 HWBOT Prime. You got 6650+ without any proof. 3rd thing, for me in fact GPUPI is pretty close (and on some cpus EASIER/runs higher!) than HWBOT Prime. Your difference looks for a pretty big one. Please share your statistics, how many CPUs you had binned from Vietnam batches and which batches.
  5. Tweaking tertiaries gives you mostly higher frequency and stability
  6. Maybe not, new Ney's concept with seasonal ranking and hw divisions gives some hope. I found one in the land of bears & vodka
  7. Nice job. Though I always play with tertiaries on X99 I like pro - looking form and seems like your conclusions are same as mine after tests. Interesting fact - XTU scores same with Quad, Triple, Dual, Single
  8. Well, we can look @ ref 1080 and custom 1080. Founder's edition gets around 2250-2300 MHz on LN2 after mods and XOC bios. Non ref 1080 can go 2400-2600 for 3D (depends from the benchmark), and more for GPU PI. So non ref 1080 TI, with proper XOC BIOS should do 100-150 MHz more than ref TitanX (also less units). Also XOC bioses mostly have slightly better efficiency, especially with high MHz. So I bet 1080 TI will beat Titan XP, but it might be lower difference than Titan X Maxwell vs 980 TI
  9. "someone please beat this and make me quit, my wife would be very grateful. " - no one will beat it cause no one will buy 4x Titan X Pascal cards You are the only one ^^
  10. Say hi to Jasmine and she is right. I have around 200 litres in dewars and it's enough
  11. Are you joining beer & retirement club with me? Great price! If I had garage I would just come with my car to grab those from you, sadly I live on 11th floor ^^
  12. Asus, Asrock all boards are fluctuating so much. Even at 102.9x/103 XTU 742/743 bunnyextraction once you have 4020 MHz cpu, once below 4 GHz. This is stable BCLK? Regarding SPI32M and wrapper - idiotic idea, why? Cause all benchmarks with wrappers are mostly broken. And SPI32M is one of the few reliable and fun benchmarks. In fact Allen you started lc results with higher cache over the limit and it was fine when you were 1st in LC challenge Just observing. I don't see anything less fun now in LC than it was in the past, especially when Asrock boards cheated on BCLK which is well known fact in past generations. Just adapt to fluctuating BCLK and that's all, find your proper golden settings and go. I don't see the difference of fun in fact when you run 100.00 or 100.20 MHz, for me SPI32M is same fun and same behaviour so I don't get the problem in fact. If Asrock fluctuates worse than Asus, maybe change board to Asus. For XTU 742/743 we have to use specific boards, best boards for that hw combination. With Skylake SPI32M LC it might be the board which fluctuates the most lol. Life. You are playing the best hw possible for the benchmark (if you can). I don't see anything to fix now, cause it's not broken. If Asrock for you is too slow/doesn't fluctuate enough, you can try M8I and do LC. The only thing which might be a good one - creating a rule that both cache and cores have to be below 5003 limit, cause it's MUCH HARDER or nearly impossible to catch both when you are running really high BCLK.
×
×
  • Create New...